Singapore – Everpure has introduced a set of data management capabilities that, according to the company, can speed up AI response times by up to 20 times as it targets the data bottlenecks holding enterprise AI back from production. They cover data governance, storage performance and AI cost control.
Everpure says the features address three problems: fragmented context, unpredictable inference costs, and slow, complex deployments. The release also builds on its “Data Primacy” principle, introduced in June, which holds that data, rather than applications, should drive enterprise architecture in the AI era.
Prakash Darji, General Manager, Data & Digital Experience of Everpure, outlined the company’s reasoning for the update. “Enterprise AI is hitting a wall not because the models are lacking, but because data is not ready for real-time, autonomous agents,” said Darji.
“We are eliminating that friction. By making enterprise data continuously governed, automated, and instantly accessible, we’re giving organisations the foundation to move AI out of the lab and into production with the necessary confidence,” Darji added.
On the governance side, Everpure Data Intelligence discovers, classifies and contextualises enterprise information at its source. It spans the Everpure Platform, public clouds, SaaS applications and third-party storage.
Building on this, Everpure has added native integration with the open Model Context Protocol (MCP). The company says AI agents and security tools can now query live data catalogues in natural language, finding relevant data and identifying its sensitivity class without custom API work.
A turnkey deployment option also runs through the existing Pure1 console. Everpure says this removes the need for professional services engagements and for separate management servers.
Another addition, privacy-first file intelligence, shows who can access each file share and how stale it is without reading file content. According to Everpure, teams can use it to fix exposure and reclaim capacity before opening shares to AI agents.
Turning to performance, Everpure has introduced PureKVA, a key-value accelerator for its FlashBlade systems. The tool pre-stages context directly into GPU memory, and the company claims it delivers up to 20x faster Time to First Token (TTFT).
Everpure adds that PureKVA supports enterprise multi-tenancy without relocating datasets. It also says the approach reduces GPU idle time, raises token throughput and cuts response lag for real-time applications.
Alongside this, the company has launched always-on DeepReduce data compression for FlashBlade. The feature scans storage blocks continuously for sub-block similarities that traditional deduplication misses, even in pre-compressed content.
Everpure says usable capacity expands automatically, with no impact on write performance and no manual scheduling. It adds that this reduces hardware footprint and cross-cloud expenses.
On cost, Everpure now offers an Intelligent Token Optimisation Reference Architecture built on open-weight models. The company says it gives organisations more control over their data and more predictable AI costs by cutting token usage from external API providers.
Everpure also links the updates to cyber resilience. Because the capabilities classify which data is sensitive and who accesses it, the same context determines how data is protected and what is recovered first, the company says.
